Puppeteer Alternatives: Playwright, Selenium, Python and Firefox Options

Which tool covers which need:

Sources Playwright for Python, Playwright browsers, Selenium Python API, Puppeteer supported browsers.

The short rule: new Python project, any browser: Playwright. The real Firefox a user has, or an existing Selenium Grid: Selenium. An existing Node.js Puppeteer project that only needs Firefox: stay on Puppeteer and launch it with browser: 'firefox'; more in what Puppeteer is.

Playwright for Python with Firefox

pip install playwright
playwright install firefox

from playwright.sync_api import sync_playwright

with sync_playwright() as p:
    browser = p.firefox.launch(proxy={"server": "http://gate.proxyfield.io:7000",
                                      "username": "<login>-country-us-city-new_york",
                                      "password": "<password>"})
    page = browser.new_page()
    page.goto("https://api.ipify.org?format=json")
    print(page.inner_text("body"))   # the exit IP, in New York
    browser.close()

Swap p.firefox for p.chromium or p.webkit and the rest of the script stays the same. For tests, pip install pytest-playwright adds a pytest plugin with a page fixture and expect() assertions.

Selenium for Python with the installed Firefox

# pip install selenium   (Selenium Manager downloads geckodriver if it is missing)
from selenium import webdriver

options = webdriver.FirefoxOptions()
options.add_argument("-headless")
options.set_preference("network.proxy.type", 1)               # manual proxy settings
options.set_preference("network.proxy.http", "gate.proxyfield.io")
options.set_preference("network.proxy.http_port", <port>)      # the port you set up in the dashboard for your allowlisted IP
options.set_preference("network.proxy.ssl", "gate.proxyfield.io")
options.set_preference("network.proxy.ssl_port", <port>)

driver = webdriver.Firefox(options=options)
driver.get("https://api.ipify.org?format=json")
print(driver.find_element("tag name", "body").text)
driver.quit()

Firefox preferences take a host and a port but no username or password, so this setup authorizes your machine's IP through an IP allowlist in the ProxyField dashboard and connects to the port you set up there (<port> above). More options, including SOCKS5: Firefox proxy setup and the Selenium integration.

Puppeteer is a solid browser library, but its scope is narrow by design (what Puppeteer is):

• Node.js only. Teams in Python, Java or .NET need another tool; Pyppeteer, the Python port, is unmaintained.
• Chrome and Firefox only. No WebKit, so no check of how Safari's engine renders a page.
• No test runner. Assertions, parallel runs and reports come from other libraries.
• Default automation signals. Out of the box, an automated Chrome differs from a person's Chrome in ways sites can see, and some sites slow or block it.
• Brittle selectors. Scripts that click by CSS path break when the layout changes, which pushes some teams toward AI-driven tools.

Versions and licenses from npm and PyPI.

Playwright keeps Puppeteer's model (browser, context, page) and adds three browser engines, four languages, locators that wait by themselves and Playwright Test. Most Puppeteer scripts port call by call: page.goto, page.click and page.evaluate keep their names. If you want one alternative and do not know which, start here. Full comparison: Puppeteer vs Playwright; background: what Playwright is.

Selenium drives browsers through WebDriver, the W3C standard every major browser vendor implements, and has bindings in six languages. Since version 4.6 it ships Selenium Manager, which downloads the right driver when it is missing (Selenium docs). It is slower to write than Playwright and needs explicit waits, but it runs the browsers people actually install, and Selenium Grid spreads tests over many machines. See what Selenium is and web scraping with Selenium and Python.

WebdriverIO is a Node.js test framework on WebDriver and WebDriver BiDi, with mobile testing through Appium. Like Cypress, it targets testing your own apps more than scraping other sites.

Crawlee is a crawling library for Node.js and Python. It wraps Playwright and Puppeteer in browser crawlers, adds fast HTTP crawlers for pages that need no browser, and brings request queues, retries, sessions and proxy rotation in one API. If you replaced Puppeteer because your scraper grew into a crawler, Crawlee often replaces your own queue code too. How to crawl a whole site: how to crawl a website.

A group of tools changes the defaults that make an automated browser look automated, for example the WebDriver flag or certain DevTools Protocol calls. They are maintained by small teams and change often, so check each project's status before you depend on it.

• nodriver is the successor of undetected-chromedriver, by the same author: a Python library that drives a regular Chrome over the DevTools Protocol without Selenium or chromedriver. It is async, marked Alpha on PyPI and licensed AGPL, which matters if you ship it inside a product. Its start() call takes a headless option, but it is built around a normal, visible Chrome.
• Pydoll is a similar async Python library for Chromium-based browsers without a WebDriver.
• Camoufox is a Firefox-based browser with a Python package that wraps Playwright, so you write ordinary Playwright code against it (PyPI).
• rebrowser-puppeteer and rebrowser-playwright are drop-in packages that apply the rebrowser patches to Puppeteer and Playwright; you change the import and keep your code.
• puppeteer-real-browser is a community package that launches a normal, headful Chrome for Puppeteer; its last release on npm was in September 2025.
• undetected-chromedriver and its successors in SeleniumBase's "UC mode" do the same for Selenium; undetected-chromedriver itself has had no release since February 2024.

What these tools do not change: the number of requests you send from one IP, the pace of your script, and the site's rules. Many sites still rate-limit by IP and watch request patterns, and a site's terms apply whatever tool you use. Scrape public pages at a reasonable pace, follow each site's terms, and read is web scraping legal. Through ProxyField, the Acceptable Use Policy applies whatever tool you run: collecting public web data is allowed, while fraud, attacks and access to data behind logins you are not entitled to are not. How sites tell browsers apart: browser fingerprinting and scraping.

Stagehand is an open-source (MIT) SDK for AI browser agents: next to ordinary code, it lets you write steps as instructions, such as act("click the next page button"), and pull structured data with extract() against a schema you define. A language model decides which element to use, so a step survives layout changes that would break a CSS selector.

Playwright vs Stagehand is less a choice than a split of work. Playwright runs deterministic steps fast and at no per-step cost; an AI agent costs a model call per step, runs slower and can be wrong, but needs less upkeep on pages that change. A common pattern is plain Playwright code for navigation and the agent only for the brittle steps. AI-based scraping tools in general: AI web scraping.

For general browser automation across engines and languages, no single tool covers more than Playwright in 2026. "Better" depends on the job: Selenium when you need the real installed browser or Selenium Grid, Crawlee for crawls with queues and retries, a patched driver when your own tests show a target treats automated browsers differently, and an AI agent SDK for flows that change every week. For pages that need no browser at all, an HTTP client beats every browser tool on speed and traffic; see Python libraries for web scraping.

• Replacing Puppeteer one for one: Playwright.
• Python, any engine including Firefox: Playwright for Python.
• The regular installed Firefox, or Selenium Grid: Selenium, with SeleniumBase if you want helpers.
• Testing your own web app: Playwright Test, Cypress or WebdriverIO.
• Crawling thousands of pages: Crawlee, or Scrapy in Python (what Scrapy is).
• Python, regular Chrome, no WebDriver: nodriver or Pydoll, after checking their licenses and status.
• Flows that break with every redesign: an AI agent SDK such as Stagehand for the fragile steps.
• Go services: chromedp or rod.

Each tool above accepts an HTTP, HTTPS or SOCKS5 proxy, either as a launch option (Playwright, Camoufox through Playwright, Crawlee's ProxyConfiguration), a browser flag or preference (Selenium, nodriver, chromedp) or the remote browser's settings. With the ProxyField gateway the address is gate.proxyfield.io:7000 with a username and password, and <login>-country-us-city-new_york asks for an exit IP in New York. Tools that set the proxy through a Chrome flag or a Firefox preference take no username and password there, so they use the IP allowlist instead, with the port you set up in the dashboard.

# Camoufox, through its Playwright wrapper (pip install camoufox)
from camoufox.sync_api import Camoufox

with Camoufox(proxy={"server": "http://gate.proxyfield.io:7000",
                     "username": "<login>-country-us-city-new_york", "password": "<password>"}) as browser:
    page = browser.new_page()
    page.goto("https://api.ipify.org?format=json")
    print(page.inner_text("body"))

ProxyField gives each request a new IP or a sticky session from 30 s to 36 h, with unlimited concurrent threads. Residential proxies cost $2.95/GB for 1 GB and fall to $1.95/GB from 1,000 GB, with no subscription; traffic never expires, and all purchases are final. Setup guides for the main drivers: Puppeteer, Playwright, Selenium. Start with residential proxies and the Easy Trial: 1 GB for $2.95.